Divergent Mechanisms of Interaction of Helicobacter pylori andCampylobacter jejuni with Mucus and Mucins

Julie Ann Naughton,a,b Karina Mariño,c Brendan Dolan,a,b Colm Reid,d Ronan Gough,d Mary E. Gallagher,d Michelle Kilcoyne,f

Jared Q. Gerlach,f Lokesh Joshi,f Pauline Rudd,c Stephen Carrington,d Billy Bourke,a,b,e Marguerite Clynea,b

School of Medicine and Medical Science, University College Dublin, Dublin, Irelanda; Conway Institute of Biomolecular and Biomedical Science, University College Dublin,Dublin, Irelandb; NIBRT Dublin-Oxford Glycobiology Laboratory, University College Dublin, Dublin, Irelandc; Veterinary Science Centre, School of Agriculture, Food Scienceand Veterinary Medicine, University College Dublin, Dublin, Irelandd; National Children’s Research Centre, Our Lady’s Children’s Hospital, Dublin, Irelande; GlycoscienceGroup, National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Irelandf

Helicobacter pylori and Campylobacter jejuni colonize the stomach and intestinal mucus, respectively. Using a combination ofmucus-secreting cells, purified mucins, and a novel mucin microarray platform, we examined the interactions of these two or-ganisms with mucus and mucins. H. pylori and C. jejuni bound to distinctly different mucins. C. jejuni displayed a striking tro-pism for chicken gastrointestinal mucins compared to mucins from other animals and preferentially bound mucins from spe-cific avian intestinal sites (in order of descending preference: the large intestine, proximal small intestine, and cecum). H. pyloribound to a number of animal mucins, including porcine stomach mucin, but with less avidity than that of C. jejuni for chickenmucin. The strengths of interaction of various wild-type strains of H. pylori with different animal mucins were comparable, eventhough they did not all express the same adhesins. The production of mucus by HT29-MTX-E12 cells promoted higher levels ofinfection by C. jejuni and H. pylori than those for the non-mucus-producing parental cell lines. Both C. jejuni and H. pyloribound to HT29-MTX-E12 mucus, and while both organisms bound to glycosylated epitopes in the glycolipid fraction of the mu-cus, only C. jejuni bound to purified mucin. This study highlights the role of mucus in promoting bacterial infection and empha-sizes the potential for even closely related bacteria to interact with mucus in different ways to establish successful infections.

The identification of two layers of mucus in the gastrointestinaltract (1, 2) and the profiling of bacterial species that inhabit

different mucosal sites throughout the body (3–5) have evokedintense interest in how bacteria colonize mucosal surfaces. Mucusis part of the innate host defense systems of humans and animals.An important function is to create a physical barrier that limitsinfection by pathogenic organisms (6). Many organisms have de-veloped strategies to overcome this barrier, and most infectionsoccur at mucosal surfaces. Understanding the molecular mecha-nisms that pathogenic organisms use to colonize the mucus layeris clearly important, since such knowledge may suggest novel ap-proaches for the prevention of colonization by pathogens.

Helicobacter pylori and Campylobacter jejuni are two closelyrelated but distinct mucosal pathogens that have adapted to infectdifferent niches in the mucus layer of the human gastrointestinaltract. H. pylori infection, one of the most common infections ofhumans, causes gastritis, duodenal ulceration, and an increasedrisk of developing gastric cancer (7). This organism colonizes thegastric mucosae of humans and some other primates but does notcolonize other hosts naturally. H. pylori predominantly colonizesthe supramucosal gel of the stomach (8). The mucus gel acts as aninfectious reservoir providing a steady supply of organisms thatcan interact with the underlying epithelium and cause disease. Thebinding of H. pylori to gastric mucin has been shown to be medi-ated via the bacterial adhesin BabA (9), which binds to the Lewisb

blood group antigen (9). H. pylori also binds to sialyl-Lewisx viathe SabA outer membrane protein (OMP) (10). H. pylori exhibitshigh genomic plasticity due to high rates of mutation and homol-ogous recombination (11, 12). Individual strains have been shownto modulate both their own adhesin expression and host glycanexpression following infection (13, 14). The dynamic modulationof H. pylori attachment is thought to confer an advantage on the

bacterium, enabling it to respond rapidly to changes in its envi-ronment.

C. jejuni is a highly motile intestinal pathogen and the leadingcause of bacterial gastroenteritis in humans in the developedworld. This organism colonizes chickens and other poultry natu-rally, without causing pathology, and chickens are an importantreservoir of infection for humans. The ability of the organism tocolonize intestinal mucus is thought to be an important virulencefactor, and exposure of C. jejuni to the human intestinal mucinMuc2 has been shown to have an effect on virulence gene expres-sion (15). Glycans present on chicken intestinal mucin have beenshown to inhibit C. jejuni invasion of HCT-8 cells (16, 17). C.jejuni can interact with blood group antigens present in humanmilk, and binding to epithelial cells is inhibited by human milk-derived fucosyl oligosaccharides (18). Recent work has profiledthe binding of C. jejuni to glycans at 42°C, the body temperature ofavian species, and 37°C, and differences have been noted (19).However, very little information exists on the carbohydrate-lectinpairs involved in mediating infections by C. jejuni.

Received 8 April 2013 Returned for modification 10 May 2013Accepted 17 May 2013

Published ahead of print 28 May 2013

Editor: B. A. McCormick

Address correspondence to Marguerite Clyne, marguerite.clyne@ucd.ie.

B.B. and M.C. contributed equally to this article and share senior authorship.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00415-13.

Copyright © 2013, American Society for Microbiology. All Rights Reserved.

doi:10.1128/IAI.00415-13

2838 iai.asm.org Infection and Immunity p. 2838–2850 August 2013 Volume 81 Number 8

New tools that enable us to study the interaction of bacteriawith mucin oligosaccharides have become available recently.These include cell lines that produce adherent mucus layers (20,21) and novel mucin microarrays that contain natural mucinsfrom different animal species (22). This study aimed to examinethe interactions of both H. pylori and C. jejuni with mucus andmucins by harnessing these novel techniques. Our results showthat these two closely related bacteria interact with mucus in verydifferent ways, and they highlight the role of mucus in promotingbacterial infection. This study also emphasizes that glycans pres-ent on host mucins and other glycosylated structures in mucuslikely play a key role in the specific species and tissue tropismdisplayed by bacteria. These data underline the importance of elu-cidating how bacteria interact with mucus in order to overcomethis important protective barrier and establish successful infec-tions.

MATERIALS AND METHODSBacterial strains, routine maintenance, and culture. The strains of C.jejuni used in this study included the well-characterized strains 81-176(23, 24) and 11168 (25), two chicken isolates, CC18 and CC19 (a gift fromSeamus Fanning, University College Dublin [UCD], Dublin, Ireland),and two human isolates, H1 and H3 (a gift from Margaret Byrne, OurLady’s Children’s Hospital, Dublin, Ireland). C. jejuni strains were rou-tinely cultured at 37°C on Mueller-Hinton agar (Oxoid) under mi-croaerophilic conditions generated using CampyGen gas packs (Oxoid).

The strains of H. pylori used in this study included the sequencedstrains 26695 (26), J99 (27), and G27 (28), as well as a babA knockoutmutant of G27, strain G27!babA (a gift from Steffen Backert). H. pyloriwas routinely cultured at 37°C on Columbia blood agar base (Oxoid)containing 7% (vol/vol) defibrinated horse blood under microaerophilicconditions generated using CampyGen gas packs (Oxoid). StrainG27!babA was cultured on plates containing 10 "g/ml of kanamycin.

Cell culture. The LS174T (29) and HT29 (30) human colonic carci-noma cell lines, methotrexate (MTX)-adapted HT29 cells (HT29-MTX)(31), and a mucus-secreting subclone, the HT29-MTX-E12 cell line (32),were used. The HT29-MTX-E12 cell line is a subclone of HT29-MTX cellsthat expresses the gastric mucin MUC5AC (21), selected on the basis oftight-junction formation and the production of an adherent mucus layer(32). HT29 cells were maintained in McCoy’s medium (Lonza) supple-mented with 10% (vol/vol) fetal bovine serum (FBS), 1% (vol/vol) non-essential amino acids (Sigma), 2 mM L-GlutaMAX (Invitrogen), 100 U/mlpenicillin (Sigma), 100 "g/ml streptomycin (Sigma), and 125 "g/ml am-photericin B (Sigma). LS174T, HT29-MTX, and HT29-MTX-E12 cellswere maintained in Dulbecco’s modified Eagle medium (DMEM; Lonza)supplemented with 10% (vol/vol) FBS, 1% (vol/vol) nonessential aminoacids (Sigma), 2 mM L-GlutaMAX (Invitrogen), 100 U/ml penicillin(Sigma), 100 "g/ml streptomycin (Sigma), and 125 "g/ml amphotericinB (Sigma).

Mucus harvesting and mucin purification. HT29-MTX-E12 cellswere grown for 21 days on Transwell filters. The mucus layer was removedfrom the cells by treatment with N-acetyl cysteine, as described previously(16, 20). LS174T cells were cultured as described above, and once theywere 60 to 80% confluent, the supernatant containing secreted mucus wascollected. Mucus samples from either the reproductive tracts or digestivetracts of cows, chickens, deer, horses, mice, sheep, pigs, and rats wereobtained either postmortem from healthy animals killed at commercialabattoirs or after humane euthanasia by an intravenous barbiturate over-dose. Generally, mucosal surfaces were scraped with a scalpel blade toharvest secreted mucus and epithelial cells. These experimental proce-dures were licensed by the Department of Health and Children, Ireland, inaccordance with the Cruelty to Animals Act (Ireland, 1897) and EuropeanCommunity Directive 86/609/EC and were sanctioned by the AnimalsResearch Ethics Committee, University College Dublin, Dublin, Ireland.

Mucins were isolated from the collected mucus and were purified as de-scribed previously (16, 22). In brief, mucus was solubilized with guanidinehydrochloride (final concentration, 4 M). Samples were reduced withdithiothreitol (DTT) (Sigma-Aldrich) at a final concentration of 0.01 Mfor 5 h at 37°C and were alkylated with iodoacetamide (0.025 M) (Sigma-Aldrich). Mucin was purified using CsCl density gradient separation fol-lowed by size exclusion chromatography (16, 22). Non-mucin-containingfractions from the CsCl gradient were retained for subsequent analysis.

Interrogation of NGC and natural mucin microarrays for bacterialbinding. Purified animal mucins and mucins from HT29-MTX-E12 andLS174T cells were printed onto microarray slides as described previously(22). Neoglycoconjugate (NGC) arrays were designed and printed as de-scribed previously (33). For incubations at 37°C, C. jejuni and H. pyloriwere cultured on agar as described above, harvested, and resuspended inMueller-Hinton broth or in brain heart infusion (BHI) broth supple-mented with 10% FBS, respectively, to an optical density at 600 nm(OD600) of 0.2. Cultures were incubated for 4 h, at which time bacteriawere harvested, washed twice in phosphate-buffered saline (PBS), andresuspended in PBS to an OD600 of 1.0.

To assess the effect of increased temperature on the binding of C.jejuni, bacteria were grown in Mueller-Hinton broth for 24 h at 37°C or42°C prior to incubation with the array. Following incubation with Syto82 (Invitrogen) for 1 h at 37°C or 42°C, bacteria were washed 7 times inlow-salt Tris-buffered saline (TBS) (22) to remove excess staining andwere resuspended to an OD600 of 1.0 in low-salt TBS with 0.05% Tween 20(TBS-T). The microarrays were incubated with fluorescently labeled bac-teria using an 8-well gasket slide and incubation cassette system (AgilentTechnologies) at 37°C or 42°C with gentle rotation in the dark. The slideswere washed 3 times with low-salt TBS-T and once in TBS. The microar-rays were dried by centrifugation and were scanned immediately.

Imaging, data extraction, and analysis. Microarray slides were im-aged in a GenePix 4000b microarray scanner (532-nm laser, 100% laserpower, 70% photomultiplier tube [PMT] setting, tetramethyl rhodamineisocyanate [TRITC] emission filter) (Molecular Devices). Data from theresultant image files were extracted using GenePix Pro, version 5.1, andwere exported as text to Excel, where all data analysis was performed.Local background was subtracted, and background-corrected median fea-ture intensity was used for each feature intensity value. For statistical use,the median of six replicate spots per subarray was treated as a single datapoint. Data were normalized to the mean for three replicate microarrayslides.

Subcellular fractionation of H. pylori and characterization of mucinbinding. Bacterial outer membrane proteins from H. pylori were lysedand isolated using sodium lauryl sarcosine as described previously (34).Briefly, H. pylori was harvested from blood agar plates, washed in sterilePBS, harvested by centrifugation at 12,000 # g for 20 min at 4°C, andfrozen at $20°C. The pellet was thawed in 20 mM Tris buffer, pH 7.5,containing protease inhibitor cocktail tablets (Roche). Following 6rounds of pulse sonication, DNase (0.1 mg/ml) and RNase (0.5 mg/ml)were added to the sonicated material, which was then incubated for 30min at room temperature (RT). Unbroken cells were removed (2,000 # g,20 min, 4°C), and the membranes were isolated by centrifugation of thesupernatant at 40,000 # g for 30 min at 4°C. The resulting pellet wasresuspended in 20 mM Tris buffer, pH 7.5, containing 0.2% sodium laurylsarcosine and was incubated at RT for 30 min. The outer membranefraction was pelleted by centrifugation at 40,000 # g for 30 min at 4°C andwas washed 3 times in ice-cold distilled water (dH2O).

Fractions were separated by electrophoresis on 10% polyacrylamideminigels and were transferred to polyvinylidene difluoride (PVDF) mem-branes overnight in a wet blotter (Bio-Rad, Hercules, CA) at 10 V. Mem-branes were blocked in 5% gelatin for 2 h at 37°C. Purified porcinestomach mucin (13) was biotinylated by using the EZ-Link sulfo-N-hy-droxysuccinimide (NHS) biotinylation kit (Pierce) according to the man-ufacturer’s instructions. Blocked membranes were incubated with 0.25mg/ml biotinylated mucin for 4 h at RT. Membranes were washed for 1 h

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2839

with TBS-Tween (0.05% [vol/vol]) and were then incubated with strepta-vidin-peroxidase (1:50,000) (Sigma) for 1 h at RT. Membranes werewashed overnight in TBS-Tween (0.05%), and bands were detected usingenhanced chemiluminescence.

Infection assays. Cells were seeded on Transwell filters (diameter, 12mm; pore size, 0.4 "m; Millipore) at a density of 1 # 105 cells/filter andwere grown for 21 days. Cells were fed every second day and were grown inantibiotic-free medium for 24 h prior to infection.

To determine the role of mucus in colonization, bacterial strains weregrown under microaerophilic conditions, and C. jejuni was prepared forinfection of the cells as described previously (20). Briefly, C. jejuni strainswere grown for 24 h on Mueller-Hinton agar plates and were then trans-ferred to a biphasic medium consisting of Mueller-Hinton agar andDMEM, supplemented with 2% FBS without antibiotics, for a further 21to 24 h. Bacteria were harvested from the biphasic medium, washed oncein DMEM, and then diluted to an OD600 of 0.2. H. pylori grown for 48 h onColumbia blood agar plates was harvested into BHI broth at pH 5.0, andthe OD600 was adjusted to 0.4.

The medium in the lower chamber of the Transwell was replaced with1.5 ml of sterile antibiotic-free McCoy’s 5A medium or DMEM, eachsupplemented with 10% (vol/vol) FBS. For C. jejuni infections, 150 "l ofa C. jejuni suspension at an OD600 of 0.2 was added to the upper chamber.H. pylori cannot survive in DMEM (35), so 100 "l of BHI broth at pH 5.0and 50 "l of the H. pylori suspension (OD600, 0.4) were added to the upperchamber. The number of organisms added to cells was approximately 1 #108 CFU/ml, which corresponded to a multiplicity of infection (MOI) of50 organisms per cell.

Infected cell cultures were incubated under microaerophilic condi-tions generated using CampyGen gas packs (Oxoid) at 37°C for 24 h.Following incubation, infected cells were washed with sterile PBS andwere then harvested using trypsin-EDTA as described previously (36).Serial dilutions of the trypsinized cells were plated out in triplicate ontoMueller-Hinton agar plates or Columbia blood agar plates for assessmentof C. jejuni or H. pylori organisms associated with the cells and wereincubated under microaerophilic conditions at 37°C. Colonies were enu-merated after 4 to 5 days of incubation.

To determine the effect of temperature on infection and invasion by C.jejuni, strains were grown for 24 h at 37°C on Mueller-Hinton agar platesand were then transferred to Mueller-Hinton broth for a further 24 h ateither 37°C or 42°C. Bacteria were harvested from the broth, washed oncein DMEM, and then diluted to an OD600 of 0.2. Cell cultures were infectedas described above and were incubated for 4 h at 37°C or 42°C. Followingthe incubation period, one set of wells was used to determine total asso-ciation as described above. The second set of wells was washed once inDMEM and was then incubated with 400 "g/ml gentamicin sulfate (Sig-ma-Aldrich) in DMEM for 2 h to kill extracellular bacteria. Cells werethen washed once in PBS and were lysed with 0.1% Triton X-100 in PBS.Serial dilutions of the cell lysate were plated out in triplicate onto Mueller-Hinton agar plates for assessment of internalized C. jejuni organisms andwere incubated under microaerophilic conditions at 37°C. Colonies wereenumerated after 3 to 4 days of incubation.

Probing of mucus and mucin of E12 cells with C. jejuni and H.pylori. Crude mucus or 10 "g of purified mucin from E12 cells was im-mobilized onto a PVDF membrane with a pore size of 0.2 "m (Millipore).Membranes were washed in PBS and were then blocked overnight in TBScontaining 5% skim milk. C. jejuni or H. pylori was cultured on agar platesas described above, harvested in PBS, washed twice, and resuspended inTBS to a final OD600 of 0.4. A suspension of either C. jejuni or H. pylori wasthen overlaid onto the membrane and was incubated at 37°C for 1 h. Theoverlays were washed three times at room temperature for 10 min on arotary shaker in TBS containing 0.05% Tween 20. Binding of C. jejuni wasdetected using a 1:5,000 dilution of a rabbit anti-C. jejuni antiserum (37)and a horseradish peroxidase (HRP)-conjugated anti-rabbit secondaryantibody (Santa Cruz). H. pylori binding was detected using an anti-H.pylori polyclonal antibody (Dako) and an HRP -conjugated anti-rabbit

secondary antibody (Santa Cruz). Reactive slots were detected using en-hanced chemiluminescence.

O-glycan analysis of purified mucin from HT29-MTX-E12 cells.Mucins were purified as described above and were freeze-dried. Then 28%NH3·H2O saturated with (NH4)2CO3 was added to a final mucin concen-tration of 1 mg/ml. The mixture was incubated at 65°C for 16 h in order torelease the O-glycans. For glycan enrichment after release, Hypercarb Mi-croSPE tips were conditioned twice with 200 "l of methanol, twice with200 "l of Milli-Q water, and finally twice with 200 "l of 10 mM ammo-nium carbonate before the sample (in an aqueous solution; final volume,200 "l) was applied. After the sample had passed through the tip, thechromatographic bed was washed 5 times with 200 "l of Milli-Q water inorder to remove excess release reagents. The retained glycans were elutedfrom the Hypercarb bed 5 times, using 100 "l of 10 mM ammoniumcarbonate solution in 80% (vol/vol) acetonitrile each time, and the eluatesfrom all these elution steps were collected in a single microcentrifuge tubeand were dried down in a SpeedVac concentrator. Once dry, the sampleswere treated with formic acid to convert them back to the aldose form andwere derivatized with 2-aminobenzamide (2-AB) as described previously(38). 2-AB-labeled glycans were analyzed by hydrophilic interaction chro-matography (HILIC) (38, 39), and the glucose unit (GU) values werecompared to those in the O-glycan database (http://glycobase.nibrt.ie/glycobase/show_nibrt.action). Further confirmation of the structuresproposed was obtained by exoglycosidase digestion.

Exoglycosidase digestion. Arrays of exoglycosidases were used incombination with HILIC– high-performance liquid chromatography(HPLC) to determine the sequence, monosaccharide type, and linkage ofsugar residues. Exoglycosidase sequencing was performed on 2-AB-la-beled O-glycans in 10 "l of a solution containing enzymes at standardconcentrations in the manufacturer’s recommended buffers for 16 h at37°C. The enzymes used were as follows: Arthrobacter ureafaciens sialidase(ABS) (EC 3.2.1.18), 1 to 2 U/ml; Streptococcus pneumoniae sialidase re-combinant in Escherichia coli (NAN1) (EC 3.2.1.18), 1 U/ml; bovine kid-ney %-fucosidase (BKF) (EC 3.2.1.51), 1 U/ml; almond meal %-fucosidase(AMF) (EC 3.2.1.111), 3 mU/ml; bovine testis &-galactosidase (BTG) (EC3.2.1.23), 2 U/ml; Streptococcus pneumoniae &-galactosidase (SPG) (EC3.2.1.23), 80 mU/ml; and &-N-acetylglucosaminidase cloned from S.pneumoniae, expressed in Escherichia coli (GUH) (EC 3.2.1.30), 10 mU/ml. After digestion, samples were separated from the exoglycosidases be-fore HPLC analysis by centrifugation in a Nanosep 10K Omega microcen-trifuge filter (Pall, VWR).

WAX chromatography. Neutral and acidic oligosaccharides were sep-arated by weak anion-exchange (WAX)-HPLC using a Vydac 301VHP575column (7.5 by 50 mm; Anachem) on a 2695 Alliance Separations modulewith a 474 fluorescence detector set (excitation and emission wavelengths,330 and 420 nm, respectively) (Waters). Solvent A consisted of 0.1 Mammonium acetate buffer, pH 7.0, in 20% (vol/vol) acetonitrile, and sol-vent B was 20% acetonitrile. Gradient conditions were as follows: a lineargradient of 0 to 5% solvent A over 12 min at a flow rate of 1 ml/min,followed by 5 to 21% solvent A over 13 min, 21 to 50% solvent A over 25min, 80 to 100% solvent A over 5 min, and 5 min at 100% solvent A.Samples were injected in water, and a fetuin N-glycan standard was usedfor calibration.

Immunofluorescent staining of HT29-MTX-E12 cells. HT29-MTX-E12 cells growing on Transwell filters were washed with PBS, removedfrom their plastic supports, and sandwiched between two thin pieces ofchicken liver prior to mounting in optimal cutting temperature (OCT)medium (BDH) as described previously, (40). Sections (20 "m) were cutusing a cryostat, collected on polylysine-coated microscope slides, al-lowed to air dry, and then either used immediately or stored at $20°Cuntil use.

Sections were fixed on slides with 2% formalin for 10 min, permeab-ilized with 0.2% (wt/vol) saponin (Sigma) in PBS for 10 min, blocked for1 h in 1% (wt/vol) bovine serum albumin (BSA; Sigma) and 10% (vol/vol)serum in PBS, and probed with an anti-Lewisb antibody (SPM194; Santa

Naughton et al.

2840 iai.asm.org Infection and Immunity

Cruz) or anti-sialyl-Lewisx (Molecular Probes) and a secondary anti-mouse antibody conjugated to Alexa Fluor 488 (Invitrogen). DAPI (4=,6-diamidino-2-phenylindole) (Invitrogen) was used to counterstain the nu-clei. Coverslips were mounted on the slides using Fluorescent MountingMedium (Dako), and the sections were examined with a fluorescencemicroscope.

Detection of Lewisb blood group antigen in HT29-MTX-E12 mucusfractions. Following CsCl gradient fractionation of E12 mucus, non-mu-cin-containing fractions were slot blotted onto a PVDF membrane (Mil-lipore). Membranes were washed in PBS and were then blocked overnightin TBS containing 5% skim milk. Lewisb was detected in the fractions byusing an antibody against Lewisb blood group antigen (SPM194; SantaCruz) and a secondary anti-mouse antibody conjugated to HRP. Reactiveslots were detected by enhanced chemiluminescence.

Glycolipid isolation and characterization and C. jejuni and H. py-lori binding. Glycolipids were isolated from mucus by multiple rounds ofchloroform-methanol extraction. Mucus was harvested as describedabove, and organic extraction of mucus was carried out according to themethod of Muindi et al. (41). Briefly, lipids were obtained by successiveextractions with chloroform-methanol (1:1 followed by 1:2 [vol/vol]) anda final extraction with chloroform-methanol-water (4.8:3.5:1). The re-sulting glycolipids were separated by thin-layer chromatography (TLC)using chloroform-methanol-0.25% KCl (5:4:1) as a solvent system withsilica TLC plates (Fisher). Glycolipid bands were visualized with a 0.1%orcinol-H2SO4 stain and were charred using a heat gun. TLC plates con-taining separated glycolipids were treated with 0.1% polyisobutylmethac-rylate in acetone, dried, and probed with antibodies against Lewisb andsialyl-Lewisx. The TLC plates were also overlaid with either H. pylori or C.jejuni organisms and were probed with antibodies raised against H. pylorior C. jejuni. TLC plates were probed with alkaline phosphatase-conju-gated secondary antibodies (Santa Cruz), and antigen-antibody com-plexes were detected using the colorimetric substrate BCIP (5-bromo-4-chloro-3-indolylphosphate)-NBT (nitroblue tetrazolium) (Sigma).

Statistical analysis. Infection assays were carried out on three separateoccasions in triplicate. Microarray interrogation was conducted in dupli-cate per microarray slide, and data were normalized to the means for threereplicate microarray slides. Graphs were drawn using GraphPad Prism.Results are presented as means ' standard deviations (indicated by errorbars) for replicate experiments. The Student t test was used to estimatestatistical significance; a P value of (0.01 was considered significant.

RESULTSThe interaction of C. jejuni and H. pylori with purified nativemucins from different animal species reveals species tropism.We investigated the binding of H. pylori and C. jejuni to a panel ofnatural mucins from different animals printed on a microarraythat enables the generation of quantitative binding data. C. jejuniand H. pylori bound to distinctly different mucins, and eachshowed tropism for mucins from specific sites among differentanimal species.

C. jejuni displayed a tropism for chicken mucin in preferenceto mucins from other animals (Table 1). All 6 isolates of C. jejuni(81-176, 11168, 2 chicken isolates, and 2 human clinical isolates)showed a clear tropism for mucin from chickens, with the strengthof the interaction dependent on the site of origin of the mucin (indescending order of interaction strength: the large intestine, prox-imal small intestine, and cecum) (Fig. 1). The binding of C. jejunito chicken mucin was not influenced by growth at human or avianbody temperature (Fig. 1). C. jejuni also bound to mucins fromother animals, including equine, porcine, ovine, and bovine mu-cins. However, the strength of the interactions with these mucinswas not as great as that seen with chicken large-intestine mucin(Table 1). The patterns of binding were consistent across different

C. jejuni isolates and were comparable at 37°C and 42°C (see TableS1 in the supplemental material).

H. pylori bound to a distinctly different subset of mucinsthan that seen with C. jejuni (Table 1). Three wild-type strainsof H. pylori and a BabA knockout strain were used to probe thearray. The wild-type organisms all displayed appreciable bind-ing to porcine stomach mucin, although the intensity of theinteraction was substantially less than the intensity of the bind-ing of C. jejuni to chicken large-intestine mucin. H. pylori alsobound to other mucins, including rat gastric and duodenalmucins (Fig. 2; Table 1). The wild-type strains all displayedsimilar binding profiles. However, strain G27!babA exhibitedreduced binding to a number of mucins, including those ofporcine and rat origins, relative to binding by the parental

TABLE 1 Binding of C. jejuni strain 81-176 and H. pylori strain 26695 tonatural mucins on a mucin microarray

Mucin sourcea

Level of bindingb of:

C. jejuni81-176

H. pylori26695

Abomasum (B) 2,780 2,158Cervicovaginal (B) 472 521Cervix (B) 497 505Duodenum (B) 795 289Spiral colon (B) 1,131 288Trachea (B) 533 261Endometrium (B) 554 177Cecum (C) 3,884 387Proximal small intestine (C) 8,775 625Large intestine (C) 14,344 571Abomasum (D) 2,021 2,261Duodenum (D) 121 487Jejunum (D) 144 653Spiral ascending colon (D) 816 172HT29-MTX-E12 cells 3,192 1,108Dorsal ascending colon (E) 925 2,036Duodenum (E) 1,183 1,794Left ventral ascending colon (E) 1,883 965Right ventral ascending colon (E) 3,255 1,786Jejunum (E) 1,891 3,064Stomach (E) 218 493Trachea (E) 313 404LS174T cells 4,298 14,409Cecum (M) 483 1,157Large intestine (M) 1,001 1,907Small Intestine (M) 202 353Stomach (M) 694 1,074Abomasum antrum (O) 343 1,126Descending colon (O) 866 1,826Duodenum (O) 1,576 1,124Ileum (O) 724 778Jejunum (O) 652 1,393Spiral colon (O) 2,654 7,759Cecum (P) 4,221 6,038Descending colon (P) 1,423 256Jejunum (P) 1,055 2,903Spiral colon (P) 1,211 3,372Stomach (P) 2,522 7,915Cecum (R) 1,031 4,302Colon (R) 1,011 482Duodenum (R) 902 5,827Ileum (R) 501 1,446Stomach (R) 1,101 4,688a B, bovine; C, chicken; D, deer; E, equine; M, mouse; O, ovine; P, porcine; R, rat.b Expressed as the mean fluorescence intensity (in relative fluorescence units) fromduplicate individual microarrays across 3 different array slides. The bindingtemperature was 37°C.

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2841

strain (Fig. 2), suggesting that the BabA outer membrane pro-tein (OMP) was playing a role in mediating the interactionbetween H. pylori G27 and a number of mucins.

Role of OMPs in mediating the binding of H. pylori to mucin.In order to investigate the potential contribution of OMPs to mu-cin binding, we probed OMPs from each of the H. pylori strainswith biotinylated porcine stomach mucin. Mucin bound specifi-

cally to an 80-kDa OMP in strains G27 and J99 but not inG27!babA or 26695 (Fig. 3A). Direct binding of these H. pyloriisolates to Lewis blood group antigen structures on an NGC arrayconfirmed that strain 26695 demonstrated lower levels of bindingto both Lewisb and sialyl-Lewisx structures than strains J99 andG27 (Fig. 3B). The results suggest that BabA, when expressed, canmediate the binding of H. pylori to porcine stomach mucin. How-ever, in the case of strain 26695, for which no OMP could bedetected to mediate binding, this interaction may be mediated bya non-protein-based bacterial adhesin.

Mucus promotes infection by both H. pylori and C. jejuni.These experiments point to a fundamental role of mucus in C.jejuni and H. pylori infection. In order to explore this further, weexamined the interactions of the organisms with HT29-MTX-E12cells, which harbor an adherent mucus layer, and compared col-onization efficiencies with parental HT29-MTX (mucus-secret-ing) and HT29 (non-mucus-secreting) cells. To determine theeffects of secreted mucins and an adherent mucus layer on theinteractions between host cells and C. jejuni or H. pylori, a com-parison was made of the colonization of 3 cell lines (HT29, HT29-MTX, and HT29-MTX-E12) with the two organisms. All coloni-zation studies were carried out on cells grown on Transwell filtersfor 21 days, because at this time the HT29-MTX-E12 cells hadproduced a mature adherent mucus layer and had formed tightjunctions, while HT29-MTX cells were also fully differentiatedand secreted mucins (21). Higher numbers of H. pylori and C.jejuni bacteria colonized the HT29-MTX-E12 cells than theHT29-MTX or HT29 cells (Fig. 4). Strikingly, H. pylori did notcolonize the non-mucus-secreting HT29 cells. In addition, whilethe organism could colonize the mucin-secreting HT29-MTXcells (6.32 # 102 CFU/ml), the number of organisms recoveredfrom the HT29-MTX-E12 cells, with an adherent mucus layer,was markedly higher (2.4 # 107 CFU/ml) (P ( 0.01) (Fig. 4). Thenumbers of C. jejuni organisms recovered from the HT29 andHT29-MTX cells (9 # 106 CFU/ml and 1.5 # 107 CFU/ml, respec-

FIG 1 Interaction of C. jejuni with chicken mucins printed on a mucin microarray slide. Microarrays containing purified chicken mucins from different sites inthe intestinal tract were probed with 6 strains of fluorescently labeled C. jejuni organisms. For incubations at 37°C, C. jejuni strains were cultured on agar asdescribed in Materials and Methods, harvested, and resuspended in Mueller-Hinton broth supplemented with 10% FBS. Cultures were grown for 4 h, at whichtime bacteria were harvested, washed, and resuspended in PBS to an OD600 of 1.0. To assess the effect of temperature on the binding of C. jejuni, bacteria weregrown in Mueller-Hinton broth for 24 h at 37°C or 42°C. Labeling and microarray interrogations were carried out at 37°C or 42°C. The histogram shows the meanfluorescence intensities from duplicate subarrays on three replicate microarray slides; values for each subarray are medians for six feature replicates. Error barsindicate the standard deviations of the means for three microarray slides. RFU, relative fluorescence units.

FIG 2 Interaction of H. pylori with porcine and rat mucins. Mucin microar-rays were probed with fluorescently labeled H. pylori organisms. Three wild-type strains and strain G27!babA were used. G27!babA does not express theBabA outer membrane protein, which mediates binding to the Lewisb bloodgroup antigen. H. pylori was cultured on agar as described in Materials andMethods, harvested, and resuspended in brain heart infusion broth supple-mented with 10% FBS. Cultures were incubated at 37°C for 4 h, at which timebacteria were harvested, washed, and resuspended in PBS to an OD600 of 1.0.Strain G27!babA was cultured in the presence of 10 "g/ml of kanamycin. Thehistogram shows the mean fluorescence intensities from duplicate subarrayson three replicate microarray slides; values for each subarray are medians forsix feature replicates. Error bars indicate the standard deviations of the meansfor three microarray slides.

Naughton et al.

2842 iai.asm.org Infection and Immunity

tively) were also higher than the numbers of H. pylori organismsrecovered from these cells. Levels of colonization of HT29 andHT29-MTX-E12 cells were comparable at 42°C and 37°C, an in-dication that the different body temperatures of humans andchickens do not influence mucus interaction (see Fig. S1 in thesupplemental material).

Binding of C. jejuni and H. pylori to HT29-MTX-E12 mucusand mucin. In light of the distinctly different mucin-binding sig-natures of H. pylori and C. jejuni, we explored further their inter-actions with purified mucus and mucin from HT29-MTX-E12cells. An antibody was used to detect bacteria on either mucus orpurified mucin from HT29-MTX-E12 cells immobilized on aPVDF membrane. C. jejuni was found to bind to both the mucusand mucin fractions. In contrast, H. pylori bound directly only tomucus; no appreciable binding to mucin was detected (Fig. 5A).Binding could be abolished by sodium metaperiodate treatmentof the mucus or mucin (Fig. 5B), indicating that the organismswere adhering to glycan epitopes. Probing of HT29-MTX-E12mucin printed on microarray slides also demonstrated markedlylower levels of binding by H. pylori than by C. jejuni (Fig. 5C).

The observation that H. pylori bound poorly to HT29-MTX-

FIG 3 Role of OMPs in mediating the binding of H. pylori to mucins. (A) Detection of H. pylori surface adhesins involved in mucin binding. Outer membraneprotein fractions of H. pylori were prepared and were probed with biotinylated porcine stomach mucin. A protein of )80 kDa, corresponding to the expected sizeof the surface adhesin BabA, was detected in strains J99 and G27 but was absent from 26695 and from a babA knockout mutant of G27. (B) Binding of H. pyloristrains to Lewis blood group antigen structures. Microarrays containing neoglyconjugate Lewis blood group antigen structures were probed with fluorescentlylabeled H. pylori organisms. The structures probed were Lewisx-BSA (LexBSA), di-Lewisx-aminophenylethyl-human serum albumin (DiLexHSA), tri-Lewisx-HSA (TriLexHSA), 3=-sialyl-Lewisx-BSA (SLexBSA), 3-sulfo-Lewisx-BSA (3SuLexBSA), 6-sulfo-Lewisx-BSA (6SuLexBSA), and lacto-N-difucohexaose I-BSA,(LNDHIBSA), which consists of Lewisb tetrasaccharide and lactose and is used as a Lewisb active structure in immunochemical and functional inhibition studies.Shown are histograms representing the mean fluorescence intensities from three replicate microarray slides of H. pylori strains binding to printed neoglycocon-jugates. The dashed line marks the level of marginal or background binding. Error bars represent the standard deviations of the means for three replicatemicroarray slides.

FIG 4 Infection of HT29, HT29-MTX, and HT29-MTX-E12 cells by mucosalpathogens. The HT29-MTX subclone E12, with an adherent mucus layer, mu-cus-secreting HT29-MTX cells, and non-mucus-secreting HT29 cells weregrown for 21 days on Transwell filters and were subsequently infected with C.jejuni or H. pylori organisms for 24 h. C. jejuni infected all cell lines. Thenumbers of C. jejuni organisms colonizing HT29-MTX and HT29-MTX-E12cells were statistically significantly higher than the number colonizing HT29cells (P ( 0.05). In contrast, H. pylori colonized HT29-MTX-E12 cells and, toa much lesser extent, HT29-MTX cells but not HT29 cells. The data presentedare means ' standard deviations for 3 replicate experiments (n * 9).

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2843

E12 mucin was surprising given that the adherent mucus layer ofthis cell line contains the gastric mucin MUC5AC (21). H. pyloridoes not colonize the colon or adhere in vivo to colonic cells.Therefore, in order to determine whether the lack of binding of H.pylori to HT29-MTX-E12 mucin was a result of the colonic originof the mucin, we compared the binding of H. pylori and C. jejuni tomucin purified from another colorectal adenocarcinoma cell line,LS174T. In contrast to the findings with HT29-MTX-E12 mucin,H. pylori interacted strongly with the LS174T mucin. In fact, C.jejuni also bound better to LS174T mucin than to E12 mucin (Fig.5C and D).

O-glycan profile of purified secreted mucin from HT29-MTX-E12 cells. Bacterial binding to mucins is mediated by in-teractions between bacterial adhesins and O-glycans on themucins. The failure of H. pylori to interact with HT29-MTX-E12 mucin suggested that the relevant fucosylated O-glycanstructures were not present on mucin originating from thesecells. In order to investigate this finding further, we determinedthe O-glycan profile of purified secreted mucin from HT29-MTX-E12 cells. The structure of O-glycans detected in HT29-MTX-E12 mucins (Table 2) was not dissimilar to those re-ported for mucins secreted from related cell lines, such as HT29(42) and HT29-MTX (43). They comprised mainly core 1- andcore 2-related structures, with a high proportion of truncatedstructures (lower GU values) (Fig. 6). Analysis of O-glycans byWAX chromatography showed that a substantial proportion ofthe O-glycans present in mucins secreted from HT29-MTX-

E12 cells were acidic. The same analysis after sialidase digestionshowed no remnant acidic structures, confirming that all thecharged species observed are sialylated glycans (Fig. 7). A no-ticeable paucity of fucosylated structures was detected, suggest-ing that a lack of relevant receptors could explain the weakinteraction of H. pylori with E12 purified mucin.

Presence of Lewisb and sialyl Lewisx antigens in HT29-MTX-E12 cells. Despite the lack of detection of fucosylated structures inmucin purified from HT29-MTX-E12 cells, staining of the cellswith an antibody against Lewisb clearly demonstrated the presenceof Lewisb blood group antigen in the adherent mucus layer of thecells. Sialyl-Lewisx staining of the mucus layer was weaker, butcellular staining for this antigen showed the presence of this struc-ture on HT29-MTX-E12 cells (Fig. 8).

Interactions of H. pylori and C. jejuni with non-mucin-con-taining fractions of E12 mucus. To determine the nonmucincomponent(s) of HT29-MTX-E12 mucus that contained Lewisb

blood group antigens, mucus fractions were obtained from aCsCl gradient and were probed with an antibody againstLewisb. The first two fractions collected from the top of thegradient (densities, 1.27 to 1.28 g/ml) reacted with the anti-Lewisb antibody (Fig. 9A). Cellular glycolipids are known to actas receptors for the binding of both H. pylori and C. jejuni (14,44) and have been found in CsCl density gradient fractions inthe range in which we found Lewisb structures (45). Therefore,we asked whether HT29-MTX-E12 mucus contained glycolip-ids and, if so, whether these could provide a source of receptors

FIG 5 Interactions of C. jejuni and H. pylori with mucins from the mucus-secreting cell lines HT29-MTX-E12 and LS174T. (A) Mucus and purified mucin fromHT29-MTX-E12 cells were immobilized on a PVDF membrane and were probed with C. jejuni strain 81-176 or H. pylori strain 26695. Both C. jejuni and H. pyloribound to immobilized mucus, but only C. jejuni bound directly to purified mucin. (B) Oxidation of the glycan structures by sodium metaperiodate reduced thebinding of C. jejuni and H. pylori to mucus as well as the interaction of C. jejuni with purified E12 mucin. (C and D) Interrogation of mucin microarrays with C.jejuni and H. pylori showed differential binding of both pathogens to mucins from two colonic cell lines, HT29-MTX-E12 (C) and LS174T (D). Histograms showmean fluorescence intensities from duplicate subarrays on three replicate microarray slides; values for each subarray are medians for six feature replicates. Errorbars are standard deviations of the means for 3 experiments.

Naughton et al.

2844 iai.asm.org Infection and Immunity

for bacterial binding. Orcinol staining demonstrated the pres-ence of glycolipids in the organic extract of the mucus sepa-rated on TLC plates (Fig. 9B). Probing of these glycolipids withantibodies against both Lewisb and sialyl Lewisx revealed thatboth glycans were expressed on the glycolipids (Fig. 9C). Fur-

thermore, overlay assays with H. pylori and C. jejuni revealedthat the organisms could bind to glycolipids isolated fromHT29-MTX-E12 mucus (Fig. 9D). These results highlight theimportance of non-mucin-based glycan interactions in medi-ating bacterial colonization of mucus.

TABLE 2 Structures of O-glycans released from E-12 mucins, described by HILIC-HPLC and exoglycosidase digestion arrays

Peak no. GUa Structureb Enzymatic digestion(s) (GU)c

1 1.00 Monosaccharides (galactose)2 1.73 BTG (partial)

3 1.99 ND BTG digest (1.00) (proposed: contaminant lactose)4 2.23 ABS (1.00), NAN1 (1.00)

5 2.41/2.48/2.54 ND Not digestible by enzymes6 2.78 GUH, no BTG

7 2.88 GUH (1.73)

8 2.99 ABS (1.73), NAN1 (1.73)

9 3.04 ND Not digestible by enzymes10 3.24 ABS (1.73), no NAN1

11 3.44 GUH (2.78)

12 3.60 BTG (2.78)

13 3.71 ABS (2.78), NAN1 (2.78)

14 4.24 GUH (3.60), GUH + SPG (2.73), GUH + BTG (2.73)

15 4.34 ABS (3.58, partial), NAN1 (3.58, partial), ABS + BTG (2.78, 3.44), ABS +SPG (2.78), no ABS + BTG + BKF, no ABS + BTG + AMF

16 4.4 ABS (1.73), NAN1 (3.24)

17 4.85 GUH (3.58), GUH + BTG (2.78)

18 5.03* BTG (4.17), BTG + GUH (2.73) (no ABS)

19 5.21* BTG (3.44), BTG + GUH (2.78)

20 5.33 ABS (3.58), NAN1 (3.58), BTG (2.78)

21 5.73* BTG (4.85), GUH + BTG (2.78)

22 6.33* BTG (4.85), BTG + GUH (2.78)

23 6.44* SPG (5.6), BTG (5.6), BTG + GUH (2.78)

a Glucose unit (GU) values are averages of triplicates, with a standard deviation of 0.06. Asterisks indicate proposed isomeric structures.b Structures are represented using the Oxford system for monosaccharides and linkages (Fig. 6, key) (59). ND, not determined.c The enzymes used were bovine testis &-galactosidase (BTG), Arthrobacter ureafaciens sialidase (ABS), Streptococcus pneumoniae sialidase (NAN1), Streptococcus pneumoniae &-N-acetylglucosaminidase (GUH), Streptococcus pneumoniae &-galactosidase (SPG), bovine kidney %-fucosidase (BKF), and almond meal %-fucosidase (AMF).

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2845

DISCUSSIONThe gastrointestinal tracts of animals and humans are rich sourcesof glycans (46, 47) due to the presence of highly glycosylated mu-cin molecules in the mucus layer overlying the epithelial cells.These glycans are often exploited by pathogens to facilitate colo-nization and disease (46, 48, 49). The expression and glycosylationof mucins differ depending on a number of factors, including thespecies, the location within the body, inflammation, and the pres-ence of microbes (48, 50, 51). To further the study of bacterialinteractions with complex glycoproteins, a mucin microarray wasdeveloped containing a wide range of natural mucins, includingthose from a number of gastrointestinal sites in several animalspecies (22). A major strength of this technique is that the binding

chemistry of the array slides allows for optimal presentation of theglycans, thereby maximizing the access of the bacteria to potentialglycan receptors. This enables a quantitative analysis of the inter-action of bacteria with secreted mucins in a high-throughput for-mat. Indeed, this improved access to potential binding sites mayexplain why we observed a degree of binding of H. pylori to mucinfrom the HT29-MTX-E12 cells on the array that was not detect-able when mucin immobilized on a PVDF membrane was used.

These natural mucin arrays have been shown previously to bean effective tool for the profiling of mucin glycoepitopes (22). Ouranalysis indicated that C. jejuni and H. pylori each interact directlywith a specific subset of mucins. Each pathogen showed a markedpreference for mucins from specific sites in certain animal species.

FIG 6 HILIC profile of fluorescently labeled O-glycans released from HT29-MTX-E12 mucins. O-glycans present in this sample were structurally characterizedby HILIC-HPLC and exoglycosidase digestion patterns (see Materials and Methods). O-glycans are represented using the Oxford system for monosaccharidesand linkages (59). A complete description of all structures can be found in Table 2; structures represented in this figure correspond to major peaks in the sample,for which numbers are italicized. Peak 4 is not a natural O-glycan but a peeling product obtained due to the chemical degradation of sialylated structures, mostprobably from structure 8 (sialyl-T antigen). The scale of glucose units (GU) is based on the elution of the 2-AB-labeled glucose ladder.

FIG 7 Weak anionic-exchange chromatography of O-glycans released from HT29-MTX-E12 mucins. Separation by charge results in different fractions ofglycans, from neutral (retention time, 4.5 min) to charged (retention time, 11 to 28 min) species. The peak at 5.5 min corresponds to excess 2-aminobenzamide.The retention times of the different O-glycans are compared to those of the fetuin N-glycans, used as standards, where S1 corresponds to monosialylatedN-glycans, and S2 to S4 correspond to di-, tri-, and tetrasialylated N-glycans, respectively. (B) Prevalence of sialylated structures in the acidic fraction ofO-glycans from HT29-MTX-E12 mucins. This panel shows a zoom-in of the chromatogram in panel A (area between 11 and 30 min) before and after A.ureafaciens sialidase (ABS) digestion.

Naughton et al.

2846 iai.asm.org Infection and Immunity

While the interaction of H. pylori with gastric mucins has beenextensively studied, little is known about the mechanisms used byC. jejuni to interact with mucins from either the human or thechicken gut. C. jejuni is most commonly identified as a commensalof chickens, which are the primary source of human infection,although it has been found in other species, including cattle,swine, and sheep (52, 53). However, despite the importance of C.jejuni colonization of chickens in the transmission of the organismto humans, there are no studies examining directly the interactionof C. jejuni with chicken mucins. C. jejuni showed a marked tro-pism for mucin obtained from chicken intestine. In addition tothis species preference, C. jejuni also showed a preference for mu-cin from specific sites in the chicken gut (in descending order ofpreference: the large intestine, proximal small intestine, and ce-cum). While a previous study demonstrated an interaction be-tween C. jejuni lipopolysaccharide (LPS) and rabbit intestinal mu-cus (54), this is the first study to report this tropism of C. jejuni forchicken mucin. In light of the current interest in bacteria and howthey interact with mucosal surfaces (3–5), our results highlight theimportance of selecting an appropriate species for the collection ofmucus for such studies. The finding that the binding of C. jejuni tochicken mucin is influenced by the anatomical location fromwhich the mucus was sourced echoes the results of our previousstudies showing a gradient of inhibition of C. jejuni invasivenessby chicken mucin, with large intestinal mucus and mucin havingthe most profound effect (16, 17). Coupled with recent data sug-gesting differences in the presence or accessibility of glycansand/or differing proportions of sulfated glycans in different partsof the chicken intestine (22), these data point to the fundamentalrole of mucin glycosylation in dictating C. jejuni tropism and life-style (commensal versus pathogenic).

The strong interaction of C. jejuni with chicken mucin suggests

that this is an important colonization factor for the organism.Therefore, alteration of the glycosylation profile of the chickenintestinal tract may be a viable means of decreasing the Campylo-bacter burden in chickens in order to reduce the risk of transmis-sion to humans. In support of this, previous studies have shownthat constituents of poultry feed alter mucin carbohydrates of thechick intestinal tract and that these alterations result in reducednumbers of C. jejuni bacteria colonizing the chicks (55).

Natural infection with H. pylori occurs only in humans andnonhuman primates. Perhaps, therefore, it is not surprising thatthe interactions between H. pylori and mucins from different an-imals were not as strong as that between C. jejuni and chickenmucin. It is of interest that H. pylori bound best to mucin isolatedfrom porcine stomach mucus, since the gnotobiotic piglet was oneof the first animal species to be experimentally infected with H.pylori (56). The BabA adhesins of H. pylori strains J99 and G27were demonstrated to interact with porcine stomach mucin.However, strain 26695, though able to interact with porcine stom-ach mucin, did not appear to have a functional BabA adhesin, asassessed both by the mucin overlay blotting experiment and bytesting of direct binding to an NGC array. This finding is in agree-ment with other studies (12). While the predicted protein se-quences of the BabA adhesins are )91% identical (using BLASTalgorithms from GenBank and the Institute for Genomic Re-search, now the J. Craig Venter Institute) for all three strains, thegene locations differ (11). In strains G27 and J99, babA is located atwhat is termed the babA locus, while in 26695, it is at a locusnormally associated with a BabA paralogue, BabB (11). Despitethe lack of a functional BabA protein, 26695 did not show a reduc-tion in binding to mucins. This is in contrast to strain G27!babA,which had significantly lower levels of binding to mucins than thewild-type strain G27. The BabA mutant was not complemented,and therefore, the possibility that spontaneous mutation or phasevariation played a role in the reduction in binding cannot be ruledout. Nevertheless, these findings suggest that while BabA is animportant H. pylori adhesin, other adhesin-glycan interactionsmust also occur (as in the case of 26695) to fulfill a similar func-tion. The binding of H. pylori to sialyl-Lewisx is mediated by theOMP SabA (10). However, strain 26695 showed low binding tosialyl-Lewisx on the NGC arrays. A previous study showed thatstrain 26695 contains predominately out-of-frame sabA allelesthat do not express functional SabA protein (57). What adhesinsmediate the binding of strain 26695 to mucin has yet to be eluci-dated.

We have shown previously that HT29-MTX-E12 cells can beinfected by both H. pylori and C. jejuni (20, 21). In this study, C.jejuni infection was enhanced in, but not dependent on, the pres-ence of mucus, whereas H. pylori required the presence of theadherent mucus layer for effective infection. This suggests thateither the mucus layer on HT29-MTX-E12 cells offered additionalH. pylori receptors not present on the other cell lines or the mucusinduced an alteration in adhesins produced by the organism thatenabled interaction with the cells. Gastric mucins have beenshown to promote the proliferation of H. pylori and to alter geneexpression by the organism (58). Recently, we have shown that theinteraction of H. pylori with the trefoil peptide TFF1, present inthe adherent mucus layer of HT29-MTX E12 cells, promotes in-fection (21). This finding highlights the role of nonmucin compo-nents of mucus in mediating the colonization of mucosal surfacesand warrants their further investigation.

FIG 8 Expression of Lewisb and sialyl-Lewisx by HT29-MTX-E12 cells. Im-munofluorescence with specific antibodies was used to assess the expression ofthe Lewisb or sialyl-Lewisx antigen by HT29-MTX-E12 cells grown for 21 dayson Transwell filters. DAPI staining was used to visualize cell nuclei. Arrow-heads indicate staining of the extracellular mucus layer, and arrows indicatecellular staining.

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2847

While C. jejuni bound strongly to purified E12 mucin, we de-tected markedly reduced interaction of H. pylori with E12 mucins.Structural analysis showed a paucity of fucosylated structures onmucin from HT29-MTX-E12 cells, despite the strong immunore-activity of the adherent mucus layer with an antibody to Lewisb

blood group antigen. CsCl gradient fractionation suggested thatthe Lewisb immunoreactive fractions colocated with glycolipids, afinding corroborated by the detection of both Lewisb and sialyl-Lewisx on purified glycolipids (Fig. 9). Moreover, both H. pyloriand C. jejuni could interact with these glycolipid fractions. Thesefindings emphasize the importance of considering mucus-boundglycolipids as a reservoir of microbial receptors that can be ex-ploited by mucosal pathogens.

In summary, this study highlights the role mucus can play inpromoting bacterial infection. It demonstrates that two closelyrelated organisms interact very differently with mucins from var-ious sources, most likely reflecting the site-specific nature of gly-cosylation. Further analysis of the glycans present in mucus maylead to the identification of unique ligands for these interactions

and, subsequently, to the development of novel therapeutic orprophylactic compounds to reduce the burden of infection.

ACKNOWLEDGMENTSThis work was supported by a grant from Science Foundation Ireland(08/SRC/B1393).

We thank Marian Kane for help with this work and also Steffen Back-ert for donating H. pylori isolates G27 and G27!babA.

REFERENCES1. Johansson ME, Phillipson M, Petersson J, Velcich A, Holm L, Hansson

GC. 2008. The inner of the two Muc2 mucin-dependent mucus layers incolon is devoid of bacteria. Proc. Natl. Acad. Sci. U. S. A. 105:15064 –15069.

2. Phillipson M, Johansson ME, Henriksnäs J, Petersson J, Gendler SJ,Sandler S, Persson AE, Hansson GC, Holm L. 2008. The gastric mucuslayers: constituents and regulation of accumulation. Am. J. Physiol. Gas-trointest. Liver Physiol. 295:G806 –G812.

3. Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI, Knight R.2009. Bacterial community variation in human body habitats across spaceand time. Science 326:1694 –1697.

FIG 9 Interactions of H. pylori and C. jejuni with the non-mucin-containing fractions of HT29-MTX-E12 cell mucus. (A) Mucus from HT29-MTX-E12 cells wasseparated on a CsCl gradient and was then divided into 1-ml fractions. A sample of each non-mucin-containing fraction was blotted onto a PVDF membrane,which was probed with an antibody against Lewisb. Shown are 3 fractions taken from the top of the density gradient. (B) TLC separation and orcinol staining ofglycolipids extracted from secreted HT29-MTX-E12 mucus. (C) TLC-separated glycolipids were probed with antibodies against Lewisb (Le B) and sialyl-Lewisx

(Sialyl Le X). (D) TLC plates containing glycolipids were probed for the binding of either C. jejuni strain 81-176 or H. pylori strain 26695 or J99. When identicalTLC plates that were not overlaid with microorganisms were probed with either anti-H. pylori or anti-C. jejuni and secondary antibodies, no signals were detected.

Naughton et al.

2848 iai.asm.org Infection and Immunity

4. Goodman AL, Kallstrom G, Faith JJ, Reyes A, Moore A, Dantas G,Gordon JI. 2011. Extensive personal human gut microbiota culture col-lections characterized and manipulated in gnotobiotic mice. Proc. Natl.Acad. Sci. U. S. A. 108:6252– 6257.

5. Muegge BD, Kuczynski J, Knights D, Clemente JC, Gonzalez A, Fon-tana L, Henrissat B, Knight R, Gordon JI. 2011. Diet drives convergencein gut microbiome functions across mammalian phylogeny and withinhumans. Science 332:970 –974.

6. Corfield AP, Carroll D, Myerscough N, Probert CS. 2001. Mucins in thegastrointestinal tract in health and disease. Front. Biosci. 6:D1321–D1357.

7. Yamaoka Y. 2010. Mechanisms of disease: Helicobacter pylori virulencefactors. Nat. Rev. Gastroenterol. Hepatol. 7:629 – 641.

8. Thomsen LL, Gavin JB, Tasman-Jones C. 1990. Relation of Helicobacterpylori to the human gastric mucosa in chronic gastritis of the antrum. Gut31:1230 –1236.

9. Van de Bovenkamp JH, Mahdavi J, Korteland-Van Male AM, BullerHA, Einerhand AW, Boren T, Dekker J. 2003. The MUC5AC glycopro-tein is the primary receptor for Helicobacter pylori in the human stomach.Helicobacter 8:521–532.

10. Mahdavi J, Sonden B, Hurtig M, Olfat FO, Forsberg L, Roche N,Angstrom J, Larsson T, Teneberg S, Karlsson KA, Altraja S, WadstromT, Kersulyte D, Berg DE, Dubois A, Petersson C, Magnusson KE,Norberg T, Lindh F, Lundskog BB, Arnqvist A, Hammarstrom L, BorenT. 2002. Helicobacter pylori SabA adhesin in persistent infection andchronic inflammation. Science 297:573–578.

11. Kawai M, Furuta Y, Yahara K, Tsuru T, Oshima K, Handa N, TakahashiN, Yoshida M, Azuma T, Hattori M, Uchiyama I, Kobayashi I. 2011.Evolution in an oncogenic bacterial species with extreme genome plastic-ity: Helicobacter pylori East Asian genomes. BMC Microbiol. 11:104. doi:10.1186/1471-2180-11-104.

12. Odenbreit S, Swoboda K, Barwig I, Ruhl S, Boren T, Koletzko S, HaasR. 2009. Outer membrane protein expression profile in Helicobacter pyloriclinical isolates. Infect. Immun. 77:3782–3790.

13. Cooke CL, An HJ, Kim J, Canfield DR, Torres J, Lebrilla CB, SolnickJV. 2009. Modification of gastric mucin oligosaccharide expression inrhesus macaques after infection with Helicobacter pylori. Gastroenterology137:1061–1071.e8.

14. Styer CM, Hansen LM, Cooke CL, Gundersen AM, Choi SS, Berg DE,Benghezal M, Marshall BJ, Peek RM, Jr, Boren T, Solnick JV. 2010.Expression of the BabA adhesin during experimental infection with Heli-cobacter pylori. Infect. Immun. 78:1593–1600.

15. Tu QV, McGuckin MA, Mendz GL. 2008. Campylobacter jejuni responseto human mucin MUC2: modulation of colonization and pathogenicitydeterminants. J. Med. Microbiol. 57:795– 802.

16. Alemka A, Whelan S, Gough R, Clyne M, Gallagher ME, CarringtonSD, Bourke B. 2010. Purified chicken intestinal mucin attenuates Cam-pylobacter jejuni pathogenicity in vitro. J. Med. Microbiol. 59:898 –903.

17. Byrne CM, Clyne M, Bourke B. 2007. Campylobacter jejuni adhere to andinvade chicken intestinal epithelial cells in vitro. Microbiology 153:561–569.

18. Ruiz-Palacios GM, Cervantes LE, Ramos P, Chavez-Munguia B, New-burg DS. 2003. Campylobacter jejuni binds intestinal H(O) antigen(Fuc%1, 2Gal&1, 4GlcNAc), and fucosyloligosaccharides of human milkinhibit its binding and infection. J. Biol. Chem. 278:14112–14120.

19. Day CJ, Tiralongo J, Hartnell RD, Logue CA, Wilson JC, von ItzsteinM, Korolik V. 2009. Differential carbohydrate recognition by Campylo-bacter jejuni strain 11168: influences of temperature and growth condi-tions. PLoS One 4:e4927. doi:10.1371/journal.pone.0004927.

20. Alemka A, Clyne M, Shanahan F, Tompkins T, Corcionivoschi N,Bourke B. 2010. Probiotic colonization of the adherent mucus layer ofHT29MTXE12 cells attenuates Campylobacter jejuni virulence properties.Infect. Immun. 78:2812–2822.

21. Dolan B, Naughton J, Tegtmeyer N, May FE, Clyne M. 2012. Theinteraction of Helicobacter pylori with the adherent mucus gel layer se-creted by polarized HT29-MTX-E12 cells. PLoS One 7:e47300. doi:10.1371/journal.pone.0047300.

22. Kilcoyne M, Gerlach JQ, Gough R, Gallagher ME, Kane M, CarringtonSD, Joshi L. 2012. Construction of a natural mucin microarray and inter-rogation for biologically relevant glyco-epitopes. Anal. Chem. 84:3330 –3338.

23. Hu L, Kopecko DJ. 1999. Campylobacter jejuni 81-176 associates withmicrotubules and dynein during invasion of human intestinal cells. Infect.Immun. 67:4171– 4182.

24. Korlath JA, Osterholm MT, Judy LA, Forfang JC, Robinson RA. 1985.A point-source outbreak of campylobacteriosis associated with consump-tion of raw milk. J. Infect. Dis. 152:592–596.

25. Parkhill J, Wren BW, Mungall K, Ketley JM, Churcher C, Basham D,Chillingworth T, Davies RM, Feltwell T, Holroyd S, Jagels K, KarlyshevAV, Moule S, Pallen MJ, Penn CW, Quail MA, Rajandream MA,Rutherford KM, van Vliet AH, Whitehead S, Barrell BG. 2000. Thegenome sequence of the food-borne pathogen Campylobacter jejuni re-veals hypervariable sequences. Nature 403:665– 668.

26. Tomb JF, White O, Kerlavage AR, Clayton RA, Sutton GG, Fleis-chmann RD, Ketchum KA, Klenk HP, Gill S, Dougherty BA, Nelson K,Quackenbush J, Zhou L, Kirkness EF, Peterson S, Loftus B, RichardsonD, Dodson R, Khalak HG, Glodek A, McKenney K, Fitzegerald LM, LeeN, Adams MD, Hickey EK, Berg DE, Gocayne JD, Utterback TR,Peterson JD, Kelley JM, Cotton MD, Weidman JM, Fujii C, Bowman C,Watthey L, Wallin E, Hayes WS, Borodovsky M, Karp PD, Smith HO,Fraser CM, Venter JC. 1997. The complete genome sequence of thegastric pathogen Helicobacter pylori. Nature 388:539 –547.

27. Alm RA, Ling LS, Moir DT, King BL, Brown ED, Doig PC, Smith DR,Noonan B, Guild BC, deJonge BL, Carmel G, Tummino PJ, Caruso A,Uria-Nickelsen M, Mills DM, Ives C, Gibson R, Merberg D, Mills SD,Jiang Q, Taylor DE, Vovis GF, Trust TJ. 1999. Genomic-sequencecomparison of two unrelated isolates of the human gastric pathogen He-licobacter pylori. Nature 397:176 –180.

28. Baltrus DA, Amieva MR, Covacci A, Lowe TM, Merrell DS, OttemannKM, Stein M, Salama NR, Guillemin K. 2009. The complete genomesequence of Helicobacter pylori strain G27. J. Bacteriol. 191:447– 448.

29. Tom BH, Rutzky LP, Jakstys MM, Oyasu R, Kaye CI, Kahan BD. 1976.Human colonic adenocarcinoma cells. I. Establishment and description ofa new line. In Vitro 12:180 –191.

30. von Kleist S, Chany E, Burtin P, King M, Fogh J. 1975. Immunohistol-ogy of the antigenic pattern of a continuous cell line from a human colontumor. J. Natl. Cancer Inst. 55:555–560.

31. Gouyer V, Leteurtre E, Zanetta JP, Lesuffleur T, Delannoy P, Huet G.2001. Inhibition of the glycosylation and alteration in the intracellulartrafficking of mucins and other glycoproteins by GalNAc%-O-bn in mu-cosal cell lines: an effect mediated through the intracellular synthesis ofcomplex GalNAc%-O-bn oligosaccharides. Front. Biosci. 6:D1235–D1244.

32. Behrens I, Stenberg P, Artursson P, Kissel T. 2001. Transport of lipo-philic drug molecules in a new mucus-secreting cell culture model basedon HT29-MTX cells. Pharm. Res. 18:1138 –1145.

33. Kilcoyne M, Gerlach JQ, Kane M, Joshi L. 2012. Surface chemistry andlinker effects on lectin– carbohydrate recognition for glycan microarrays.Anal. Methods 4:2721–2728.

34. Reeves EP, Ali T, Leonard P, Hearty S, O’Kennedy R, May FE, WestleyBR, Josenhans C, Rust M, Suerbaum S, Smith A, Drumm B, Clyne M.2008. Helicobacter pylori lipopolysaccharide interacts with TFF1 in a pH-dependent manner. Gastroenterology 135:2043–2054.e2.

35. van Amsterdam K, van der Ende A. 2004. Nutrients released by gastricepithelial cells enhance Helicobacter pylori growth. Helicobacter 9:614 –621.

36. Cottet S, Corthesy-Theulaz I, Spertini F, Corthesy B. 2002. Microaero-philic conditions permit to mimic in vitro events occurring during in vivoHelicobacter pylori infection and to identify Rho/Ras-associated proteinsin cellular signaling. J. Biol. Chem. 277:33978 –33986.

37. Mooney A, Byrne C, Clyne M, Johnson-Henry K, Sherman P, BourkeB. 2003. Invasion of human epithelial cells by Campylobacter upsaliensis.Cell. Microbiol. 5:835– 847.

38. Royle L, Campbell MP, Radcliffe CM, White DM, Harvey DJ, Abra-hams JL, Kim YG, Henry GW, Shadick NA, Weinblatt ME, Lee DM,Rudd PM, Dwek RA. 2008. HPLC-based analysis of serum N-glycans ona 96-well plate platform with dedicated database software. Anal. Biochem.376:1–12.

39. Merry AH, Neville DC, Royle L, Matthews B, Harvey DJ, Dwek RA,Rudd PM. 2002. Recovery of intact 2-aminobenzamide-labeledO-glycans released from glycoproteins by hydrazinolysis. Anal. Biochem.304:91–99.

40. Keely S, Rawlinson LA, Haddleton DM, Brayden DJ. 2008. A tertiaryamino-containing polymethacrylate polymer protects mucus-covered in-testinal epithelial monolayers against pathogenic challenge. Pharm. Res.25:1193–1201.

41. Muindi K, Cernadas M, Watts GF, Royle L, Neville DC, Dwek RA,

Interaction of Bacteria with Mucus and Mucins

August 2013 Volume 81 Number 8 iai.asm.org 2849

Besra GS, Rudd PM, Butters TD, Brenner MB. 2010. Activation stateand intracellular trafficking contribute to the repertoire of endogenousglycosphingolipids presented by CD1d. Proc. Natl. Acad. Sci. U. S. A.107:3052–3057. (Corrected.)

42. Huet G, Gouyer V, Delacour D, Richet C, Zanetta JP, Delannoy P,Degand P. 2003. Involvement of glycosylation in the intracellular traffick-ing of glycoproteins in polarized epithelial cells. Biochimie 85:323–330.

43. Capon C, Laboisse CL, Wieruszeski JM, Maoret JJ, Augeron C, FournetB. 1992. Oligosaccharide structures of mucins secreted by the humancolonic cancer cell line CL.16E. J. Biol. Chem. 267:19248 –19257.

44. Lingwood CA. 1999. Glycolipid receptors for verotoxin and Helicobacterpylori: role in pathology. Biochim. Biophys. Acta 1455:375–386.

45. Cheung MD, Albers JJ. 1982. Distribution of high density lipoproteinparticles with different apoprotein composition: particles with A-I andA-II and particles with A-I but no A-II. J. Lipid Res. 23:747–753.

46. Hooper LV, Gordon JI. 2001. Commensal host-bacterial relationships inthe gut. Science 292:1115–1118.

47. Sonnenburg JL, Xu J, Leip DD, Chen CH, Westover BP, WeatherfordJ, Buhler JD, Gordon JI. 2005. Glycan foraging in vivo by an intestine-adapted bacterial symbiont. Science 307:1955–1959.

48. Corfield AP, Wiggins R, Edwards C, Myerscough N, Warren BF,Soothill P, Millar MR, Horner P. 2003. A sweet coating— how bacteriadeal with sugars. Adv. Exp. Med. Biol. 535:3–15.

49. Moran AP, Gupta A, Joshi L. 2011. Sweet-talk: role of host glycosylationin bacterial pathogenesis of the gastrointestinal tract. Gut 60:1412–1425.

50. Lebeer S, Vanderleyden J, De Keersmaecker SC. 2010. Host interactionsof probiotic bacterial surface molecules: comparison with commensalsand pathogens. Nat. Rev. Microbiol. 8:171–184.

51. Singh PK, Hollingsworth MA. 2006. Cell surface-associated mucins insignal transduction. Trends Cell Biol. 16:467– 476.

52. Fricker CR, Park RW. 1989. A two-year study of the distribution of‘thermophilic’ campylobacters in human, environmental and food sam-ples from the Reading area with particular reference to toxin productionand heat-stable serotype. J. Appl. Bacteriol. 66:477– 490.

53. Zanetti F, Varoli O, Stampi S, De Luca G. 1996. Prevalence of thermo-philic Campylobacter and Arcobacter butzleri in food of animal origin. Int.J. Food Microbiol. 33:315–321.

54. McSweegan E, Walker RI. 1986. Identification and characterization oftwo Campylobacter jejuni adhesins for cellular and mucous substrates.Infect. Immun. 53:141–148.

55. Fernandez F, Sharma R, Hinton M, Bedford MR. 2000. Diet influ-ences the colonisation of Campylobacter jejuni and distribution of mu-cin carbohydrates in the chick intestinal tract. Cell. Mol. Life Sci. 57:1793–1801.

56. Eaton KA, Morgan DR, Krakowka S. 1992. Motility as a factor in thecolonisation of gnotobiotic piglets by Helicobacter pylori. J. Med. Micro-biol. 37:123–127.

57. Goodwin AC, Weinberger DM, Ford CB, Nelson JC, Snider JD, HallJD, Paules CI, Peek RM, Jr, Forsyth MH. 2008. Expression of theHelicobacter pylori adhesin SabA is controlled via phase variation and theArsRS signal transduction system. Microbiology 154:2231–2240.

58. Skoog EC, Sjoling A, Navabi N, Holgersson J, Lundin SB, Linden SK.2012. Human gastric mucins differently regulate Helicobacter pylori pro-liferation, gene expression and interactions with host cells. PLoS One7:e36378. doi:10.1371/journal.pone.0036378.

59. Harvey DJ, Merry AH, Royle L, Campbell MP, Dwek RA, Rudd PM.2009. Proposal for a standard system for drawing structural diagrams ofN- and O-linked carbohydrates and related compounds. Proteomics9:3796 –3801.

Naughton et al.

2850 iai.asm.org Infection and Immunity